To travel from one location to another is a common task for most human beings. Actually creating a robot to find a path to the destination while avoiding obstacles is another task. The behavior of an artificial neural network is synthesized mimicking the decisions leading to a path. a two-layer network using a simplified model of a biological network (i.e. brain is used).
There are two pseudo-sub-networks within the network itself: one is the proximity sub-network which attempts to travel toward the destination; the other is the ray sub-network which attempts to avoid obstacles. However, conflicts occur when avoiding an obstacle which also means deviating from the course toward these destinations. Adjustments to the priority of the two sub-network had been closer to 100%. But when the ray sub-network is also used, circumstances will arise when the robot over-corrects itself and misses the destination, thus causing it to loop around the destination. Sometimes the robot may oscillate forwards and backwards.
Although it was not possible to have a success rate of 100%, the author believes that by extending the network such as employing the switches or adding oscillators, the robot may have a completion rate closer to 100%. Such conjectures shall be tested in the future.
Upper Arlington High School, Upper Arlington, Ohio